EMRFD Message Archive 6239

Message Date From Subject
6239 2011-05-03 08:07:52 Chris High Isolation Feedback Amplifiers
I deleted the earlier file of feedback amplifier comparisons and replaced it with a revised version in the files section:

Feedback Amplifier Comparisons Rev A.pdf

This version has an additional schematic showing a reduced component approach that eliminates two capacitors and which uses a 3-winding hybrid transformer as a voltage balun in place of the earlier 2-winding power splitter. Performance is essentially the same.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
6241 2011-05-03 12:25:47 Lasse Re: High Isolation Feedback Amplifiers
Chris,
I had a peek at the new document, interesting enough for me to try my
linear simulator on that reduced component circuit!

Picked a random NPN, which turned out to be s-parameters for 2N3866, and
I had excellent numbers, even better than what you presented. So I
started to try other transistors.
BFR91 and 96 was a LOT worse. Isolation detoriated 20 dB (going from ~50
dB to 30dB! Found a file for 2N5179 which gave best isolation of the
bunch. 2N5109 somewhere in the middle.

So what parameters in the transistor do affect the isolation? It is like
a low-frequency transistor offers better performance (i.e. isolation)
than a high frequency one.

Cheers
Lasse SM5GLC

Chris skrev 2011-05-03 17:07:
>
> This version has an additional schematic showing a reduced component approach that eliminates two capacitors and which uses a 3-winding hybrid transformer as a voltage balun in place of the earlier 2-winding power splitter. Performance is essentially the same.
>
>
> Chris Trask
> N7ZWY / WDX3HLB
> Senior Member IEEE
> http://www.home.earthlink.net/~christrask/
>
6242 2011-05-03 12:45:30 Chris Trask Re: High Isolation Feedback Amplifiers
>
> I had a peek at the new document, interesting enough for me to try my
> linear simulator on that reduced component circuit!
>
> Picked a random NPN, which turned out to be s-parameters for 2N3866, and
> I had excellent numbers, even better than what you presented. So I
> started to try other transistors.
> BFR91 and 96 was a LOT worse. Isolation detoriated 20 dB (going from ~50
> dB to 30dB! Found a file for 2N5179 which gave best isolation of the
> bunch. 2N5109 somewhere in the middle.
>
> So what parameters in the transistor do affect the isolation? It is like
> a low-frequency transistor offers better performance (i.e. isolation)
> than a high frequency one.
>

Primarily for a common emitter (CE) amplifier it's the collector-to-base
capacitance (Ccb or Co). In the hybrid-pi linear model it's seen as a
single capacitor going from the collector terminal to the junction of the
base spreading resistance rbb and the base-emitter resistance rpi. A fairly
recent improvement has two capacitors, each half of Ccb, both from the
collector, one going to a midway point of rbb and the other going to the
junction as before. Seems that it comes closer to measured data in most
cases.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
6243 2011-05-03 13:50:54 Lasse Re: High Isolation Feedback Amplifiers
Sorry if I'm beeing a Simpleton,
shouldn't then a smaller Ccb offering better reverse isolation at low
frequencies??

My simulation using s-parameters shows worse performance with a high fT
transistor i.e. BFR96 Ccb2,6pF vs. 2N3866 Ccb=3pF

Cannot find a number for a 2N2222 though... only Cc of 8pF
/Lasse SM5GLC

Chris Trask skrev 2011-05-03 21:45:
>
> Primarily for a common emitter (CE) amplifier it's the collector-to-base
> capacitance (Ccb or Co). In the hybrid-pi linear model it's seen as a
> single capacitor going from the collector terminal to the junction of the
> base spreading resistance rbb and the base-emitter resistance rpi. A fairly
> recent improvement has two capacitors, each half of Ccb, both from the
> collector, one going to a midway point of rbb and the other going to the
> junction as before. Seems that it comes closer to measured data in most
> cases.
>
>
6245 2011-05-03 17:26:37 Chris Trask Re: High Isolation Feedback Amplifiers
>
> Sorry if I'm beeing a Simpleton,
> shouldn't then a smaller Ccb offering better reverse isolation at low
> frequencies??
>
> My simulation using s-parameters shows worse performance with a high fT
> transistor i.e. BFR96 Ccb2,6pF vs. 2N3866 Ccb=3pF
>
> Cannot find a number for a 2N2222 though... only Cc of 8pF
>

It should be, and I've always assumed that it was.

I just tried using a 2N3866 in the same circuit, and got a reverse
isolation of -54.4dB, output RL of -46.9dB, input RL of -26.3dB, and output
noise of 1.918nV.

The Ccb for the PSpice models I have here for the 2N2222 is 7.306pF and
for the 2N3866 is 9.05pF. Since the emitter is not grounded, it's very
likely that there is an additional reverse path through the
collector-emitter (output) capacitance and then through the base-emitter
junction.

Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
6246 2011-05-04 09:23:21 kb1gmx Re: High Isolation Feedback Amplifiers